51 karma · joined January 5, 2013
[ my public key: https://keybase.io/nikaspran; my proof: https://keybase.io/nikaspran/sigs/L8mPn-ORtSd4QbmR4mmkmKanX5ULtiPjZ-xIFSjxkP4 ]
As for using DBs, that's certainly an option (i.e. langchain and such), but at some point you do still need to bring in the data inside the context, so I'd say it's still interesting to consider what would be an efficient way to represent that data via text.
I wonder if we might also see LLM specific data serialisation formats in the future, to make use of tokenization in the most efficient manner and enhance the generative capability of the models.
It suggests subreddits based on what you're already subscribed to.
Denying entry into airspace is not the same as forcing a plane to land and kidnapping a political opponent.
I personally prefer Fusion 360 (it feels sleeker and is way more usable with a touchpad), but it's definitely a viable choice.
It's been an awesome exercise and I've been quite impressed by the API available to extensions.
I.e.: http://searchengineland.com/tested-googlebot-crawls-javascri...
=================
Hi <...>,
I'm glad that you find it amazing but we have Method on hold for now. We are focusing on other projects we think will work better.
Thanks for your kind words,
Maria.
mariamunuera.com
=================
I got this response after I sent an email asking for any status updates sometime in April. Unfortunate.